The Anatomy of Memory: an Interactive Overview of the Parahippocampal– Hippocampal Network

REVIEWS

memory systems The anatomy of memory: an interactive overview of the parahippocampal– hippocampal network

N. M. van Strien*||, N. L. M. Cappaert‡|| and M. P. Witter*§ Abstract | Converging evidence suggests that each parahippocampal and hippocampal subregion contributes uniquely to the encoding, consolidation and retrieval of declarative memories, but their precise roles remain elusive. Current functional thinking does not fully incorporate the intricately connected networks that link these subregions, owing to their organizational complexity; however, such detailed anatomical knowledge is of pivotal importance for comprehending the unique functional contribution of each subregion. We have therefore developed an interactive diagram with the aim to display all of the currently known anatomical connections of the rat parahippocampal–hippocampal network. In this Review, we integrate the existing anatomical knowledge into a concise description of this network and discuss the functional implications of some relatively underexposed connections.

In the more than 100 years since the first explorations of underexposed connections tend to be erased from the the parahippocampal–hippocampal network by Ramon common scientific memory. For this Review, we have y Cajal1, numerous detailed anatomical tract-tracing assembled the extensive anatomical PHR–HF connec- analyses (BOX 1) have been published. These studies were tivity literature, focusing on all known connections of sparked by the discovery of a prominent relationship one frequently used experimental animal: the rat. We between declarative memory and structures in the human introduce a new approach to describe the network con- medial temporal lobe, in particular the hippocampal for- nectivity that uses an interactive diagram to display the mation (HF)2; the importance of the parahippocampal complete PHR–HF connectivity (see Supplementary region (PHR) for memory was established only later3. An information S1 (figure) and Supplementary informa- *Department of Anatomy and increasingly complex picture of the connectivity within tion S2 (box)). The complex and detailed connectivity Neurosciences, VU University and between the HF and the PHR has emerged over the patterns in this diagram are made accessible through Medical Center, P.O. BOX 7057, 1007 MB years, and comprehensive knowledge of the PHR–HF the ability to switch on and off individual or groups of 4 Amsterdam, The Netherlands. network lies at the basis of understanding its functions . network connections between cortical layers and/or ‡SILS Center for The level of anatomical detail at which an experi- anatomical areas. The information this diagram pro- Neuroscience, University of ment must be carried out or results interpreted vides could prove to be useful at a time when research Amsterdam, 1098 SM depends on the questions under investigation. In is moving beyond the functional explanations that can Amsterdam, The Netherlands. §Kavli Institute for Systems some instances, the effects of experimental manipula- be provided by a PHR–HF circuitry model that contains Neuroscience, and Centre for tions can be interpreted using connectivity data at an only a subset of the connections; moreover, it might the Biology of Memory, overall network level (without taking the details of lead to a re-evaluation of the functional importance of Department of Neuroscience, local networks into account). Other studies require connections that have previously been ignored. Norwegian University of Science and Technology, more detail, but even those studies that benefit from This Review first describes the anatomical concepts N-7489 Trondheim, Norway. a detailed understanding of the circuitry often do not, that are essential to understanding the PHR–HF cir- ||These authors contributed for a variety of reasons, take all the known connections cuitry (for an extensive description, see REFS 5–7). Next, equally to this work. into consideration. Sometimes connections are sim- it presents an overview of the main PHR–HF circuits as Correspondence to N.M.v.S. ply overlooked, whereas other times connections are well as of some of the lesser-known aspects of the cir- e-mail: n.vanstrien@temporal-lobe. intentionally left out because they seem to have no cuitry, using the interactive diagram (Supplementary com function and are therefore considered irrelevant for a information S1 (figure)). Subsequently, it shows how doi:10.1038/nrn2614 particular theoretical interpretation. Eventually, such having detailed knowledge of the PHR–HF circuitry can

Box 1 | Neuroanatomical tract-tracing methods consisting of medial (mEA) and lateral (lEA) areas), the perirhinal cortex (PER, consisting of Brodmann Most of what is known today about the pathways that connect neurons in different areas (A) 35 and 36) and the postrhinal cortex (POR). brain regions has been discovered by using neuroanatomical tract-tracing The PHR is generally described as having six layers. The techniques156. A tracer is a substance that allows such pathways to be visualized. coordinate systems that define position within the HF Tracers can be injected intracellularly to label the dendrites and axons of a neuron. FIG. 1 Both autofluorescent dyes (for example, Lucifer yellow and Alexa dyes) and and the PHR are explained in . biotin-derived dyes are often used for intracellular labelling, as they can be easily visualized using fluorescent microscopy. Alternatively, a tracer can be injected at a Circuitry of the PHr–HF region stereotaxically defined extracellular location in the in vivo brain. The tracer is taken up In the interactive diagram (FIG. 2; Supplementary infor- by neurons at the injection site and is transported or diffuses within cells. A tracer mation S1 (figure)) we attempted to display all of the substance can be transported anterogradely from the soma towards the axon PHR–HF connections that have been reported in the ana- terminals (for example, Phaseolus vulgaris leucoagglutinin), retrogradely from the tomical literature concerning the rat (for references see axon terminals towards the soma (for example, Fast Blue), or it can be transported in Supplementary information S3 (table)). The interactive both directions (for example, horseradish peroxidase). Another tract-tracing method diagram contains almost 1,600 connections, which can be involves creating small lesions and visualizing the resulting degeneration; the labelled displayed at a customizable level of complexity. This allows connections are generally assessed using light microscopy. Electron microscopy can be used to visualize whether a presynaptic axon contacts a postsynaptic element. This is a easy comparisons between the detailed PHR–HF circuitry very accurate but time-consuming method because only small pieces of tissue can be illustrated by the diagram and a ‘standard’ model of this examined at one time. Alternatively, confocal microscopy allows three-dimensional circuitry (FIG. 3), which displays the subset of connections reconstruction of larger pieces of tissue and can indicate whether pre- and that are currently most often used in the field (based on an postsynaptic elements are likely to form a synapse. A question of current interest is analysis of a selection of recent key studies8–15). whether confocal microscopy is reliable enough for indicating such contacts. In order to increase our understanding of the connectivity of the brain and its related function, Connectivity within the PHR. In the standard model accurate numbers that provide information about pathways’ projection intensity and (FIG. 3), the projections from the PER and the POR to the 157 termination density are needed. To achieve this, techniques using viral tracers EC are often depicted with a topology that emphasizes and new genetic tools158 are being developed. PER-to-lEA and POR-to-mEA relationships. However, as can be seen in the interactive diagram, the available data indicate (see figure 1a in Supplementary informa- aid one’s understanding of some of the functional pro- tion S4 (figure)) that the POR also projects to the lEA, cesses that engage the PHR–HF regions, such as memory although quantitatively to a lesser extent than the PER formation, spatial navigation and temporal dynamics. (4.9% versus 15.6%, respectively, of the total cortical input)16. likewise, the PER also projects to the mEA (see Hippocampal–parahippocampal anatomy figure 1b in Supplementary information S4 (figure)), con- The rat HF is a C-shaped structure that is situated in tributing a level of cortical input equal to that of the POR the caudal part of the brain. Three distinct subregions (7.5%)16. Neurons in layers II, III, V and VI of A35 and can be distinguished (FIG. 1): the dentate gyrus (DG), A36 of the PER project in a convergent way to lEA layers the hippocampus proper (consisting of CA3, CA2 and II and III17, whereas the PER projection to the mEA arises CA1) and the subiculum. The cortex that forms the mainly from A36 (REFS 16,17). The POR projection to the HF has a three-layered appearance. The first layer is a lEA arises from layers II, III, V and VI and terminates deep layer, comprising a mixture of afferent and efferent in layers II and III16,17. The POR projection to the mEA fibres and interneurons. In the DG this layer is called originates from the same layers and terminates preferen- the hilus, whereas in the CA regions it is referred to tially in the superficial layers, although some fibres can as the stratum oriens. Superficial to this polymorph be seen in the deep layers of the mEA16,18. layer is the cell layer, which is composed of principal The EC reciprocates the projections from the PER cells and interneurons. In the DG this layer is called the and the POR, as depicted in the standard model. A granule layer, whereas in the CA regions and the subicu- detailed look at the interactive diagram shows that lum it is referred to as the pyramidal cell layer (stratum there are projections from layers III and V of the lEA to pyramidale). The most superficial layer is referred to all layers of A35 and A36 (REFS 16,17,19), and from the as the molecular layer (the stratum moleculare) in the mEA to all layers of A35 (REFS 16,17,20) (see figure 2a DG and the subiculum. In the CA region the molecular in Supplementary information S4 (figure)). The mEA layer is subdivided into a number of sublayers. In CA3, also projects to A36 (REFS 16,21). Both the mEA and the three sublayers are distinguished: the stratum lucidum, lEA project to the POR, but details of the topography which receives input from the DG; the stratum radiatum, of this connection in the rat are currently not avail- Temporal dynamics comprising the apical dendrites of the neurons located able16,21,22 (see figure 2b in Supplementary information Properties of neurons in a network, such as precise spike in the stratum pyramidale; and, most superficially, the S4 (figure)). times and firing rates, that stratum lacunosum-moleculare, comprising the apical Traditionally, little attention has been paid to the con- facilitate information transfer. tufts of the apical dendrites. The lamination in CA2 nections between the PER and the POR, although there is and CA1 is similar, with the exception that the stratum extensive connectivity between these regions. POR layers Convergence lucidum is missing. II and V project to all layers of A35 and A36; POR layer When inputs from different brain regions congregate on to The PHR lies adjacent to the HF, bordering the subic- III also projects to A36 (see figure 3a in Supplementary single cells or on to a local ulum. It is divided into five subregions: the presubicu- information S4 (figure)). Rostral levels of the POR pro- network in another region. lum, the parasubiculum, the entorhinal cortex (EC, vide the densest projection to caudal levels of A35 and

A cd Dorsal

Septal

Septal a

Temporal dl Temporal vm Ventral

Rostral Caudal Lateral Medial

B a c C

CA3 Sub CA1 CA1 DG Deep Superficial Septal or DG CA3 pyr luc rad Sub CA3 PrS slm or POR A36 PaS MEA CA1 A35 Prox VI/V Temporal Dist LEA DG rad III/II slm pyr I A36 Sub Olfactory cortex exp hilus gl encl VI/V b d ml PrS ml III/II Distal I pyr I PaS II/III A35 I II/III V/VI CA3 Sub VI PrS V VI CA1 A36 III IV Proximal I II V DG IV PrS POR Sub A35 PaS MEA III PaS II LEA MEA LEA MEA LEA I

Figure 1 | representations of the hippocampal formation and the parahippocampal region in the rat brain. a | Lateral (left panel) and caudal (right panel) views. For orientation in the hippocampal formation (consisting of the dentate gyrus (DG; dark brown), CA3 (medium brown), CA2 (not indicated), CA1 (orange) andNa thetur subiculume Reviews | Neur(Sub;oscienc yellow)),e three axes are indicated: the long or septotemporal axis (also referred to as the dorsoventral axis); the transverse or proximodistal axis, which runs parallel to the cell layer and starts at the DG; and the radial or superficial-to-deep axis, which is defined as being perpendicular to the transverse axis. In the parahippocampal region (green, blue, pink and purple shaded areas), a similar superficial-to-deep axis is used. Additionally, the presubiculum (PrS; medium blue) and parasubiculum (PaS; dark blue) are described by a septotemporal and proximodistal axis. The entorhinal cortex, which has a lateral (LEA; dark green) and a medial (MEA; light green) aspect, is described by a dorsolateral-to-ventromedial gradient and a rostrocaudal axis. The perirhinal cortex (consisting of Brodmann areas (A) 35 (pink) and 36 (purple)) and the postrhinal cortex (POR; blue-green) share the latter axis with the entorhinal cortex and are additionally defined by a dorsoventral orientation. The dashed lines in the left panel indicate the levels of two horizontal sections (a,b) and two coronal sections (c,d), which are shown in part B. All subfields of the parahippocampal–hippocampal region are colour-coded in correspondence with the interactive diagram in Supplementary information S1 (figure). A further description of the anatomical features of each subfield is provided in the legend of this supplementary information. c | A Nissl-stained horizontal cross section (enlarged from part Bb) in which the cortical layers and three-dimensional axes are marked. The Roman numerals indicate cortical layers. CA, cornu ammonis; dist, distal; dl, dorsolateral part of the entorhinal cortex; encl, enclosed blade of the DG; exp, exposed blade of the DG; gl, granule cell layer; luc, stratum lucidum; ml, molecular layer; or, stratum oriens; prox, proximal; pyr, pyramidal cell layer; rad, stratum radiatum; slm, stratum lacunosum-moleculare; vm, ventromedial part of the entorhinal cortex.

Reciprocal connections A36. Additionally, the POR projection to A36 is stronger The deep layers of the presubiculum project to all lay- Bidirectional, equivalent than that to A35 (REFS 16,17,23). The PER projection to ers of the mEA and predominantly to the deep layers of connections between two the POR originates in PER layers II, V and VI16,17,21,23 (see the lEA27,34,35. A detailed topography for the parasubic- areas, networks or neurons. figure 3b in Supplementary information S4 (figure)). ulum-to-EC connection has not yet been described, Perforant pathway The densest projection connects the rostral PER with but it is known that all layers of the parasubiculum 17 21,24,30,32,33,36 Axons that originate in the the caudal POR . converge on to layer II of the mEA . superficial layers of the EC and A set of intra-PHR connections that is also under- Several other connections have been described that are distributed to all fields of exposed in the standard model is the connections have not been incorporated into the standard model the hippocampus. between the EC, the presubiculum and the parasubicu- shown in FIG. 3. For example, reciprocal connections lum. The dorsolateral mEA projects to septal levels of between the presubiculum/parasubiculum and the the presubiculum and the parasubiculum (see figure 4a PER/POR have been described21,23, but details are lim- in Supplementary information S4 (figure)), whereas ited. Other connections, such as the intrinsic connec- the ventromedial mEA projects to the temporal pre- tions of the EC, are better anatomically characterized, subiculum and parasubiculum20,22,24–28 (see figure 4b in but they remain outside the scope of most models. For Supplementary information S4 (figure)). The lEA also example, the mEA and the lEA are strongly intercon- projects to the presubiculum and the parasubiculum, but nected: cells in layers II, III, V and VI of the mEA project precise topographical information for this projection is to the superficial layers of the lEA20,37; lEA layers II and absent19,20,22,25,29,30. Both the presubiculum and the par- V project to the superficial layers of the mEA17,20,29,37, asubiculum send projections to the EC. The septal pre- whereas lEA layers III and VI project to superficial and subiculum projects to the dorsolateral and intermediate deep layers of the mEA29,37. part of the mEA (see figure 5a in Supplementary information S4 (figure)), whereas the temporal presubiculum PHR projections to the HF. There is a prominent and projects to the ventromedial part of the mEA (see figure topologically arranged circuitry between the PHR 5b in Supplementary information S4 (figure)). The super- and the HF. The EC-to-HF circuitry is known as the ficial layers of the presubiculum project to the deep layers perforant pathway (FIG. 3). According to the standard of the lEA31 and to layers I, II and III of the mEA27,32–34. view only EC layer II projects to the entire transverse

a b

Figure 2 | interactive diagram. The interactive diagram (see Supplementary information S1 (figure)) shows the details of the connectivity in the parahippocampal–hippocampal network, including the topology of theNatur connections.e Reviews | Neur All regionsoscienc e and their three-dimensional axes (for example, the septotemporal axis; see FIG. 1) are included in the diagram. a | An alphabetically sorted list of ‘from–to’ connection groups that can be switched on or off. In front of each group is a + sign. Clicking this expands the list of individual connections that make up the group, allowing one to select connections originating from a specific cortical layer or according to a specific three-dimensional projection pattern (for example, only dorsolateral entorhinal cortex to septal hippocampus connections). b | In this area of the diagram the selected connectivity within and between subregions is displayed with full topological detail. c | In some cases topological detail is not available; these connections are displayed with a reduced level of topological detail in the centre of the diagram. Connections between diagram elements in parts b and c also exist. d | The diagram legend provides a detailed anatomical description of all subregions. Refer to the diagram manual in Supplementary information S2 (box) for detailed instructions.

Divergence extent of the DG. In fact, EC layers III, V and VI also When one brain region sends contribute to this projection, although to a lesser HF DG CA3 CA1 Sub projections to several different extent. The details of the EC-to-DG19,22,24–26,38–51 and brain regions. EC-to-CA319,20,22,25,29,38,41,42,44,48–50,52 projections might Mossy fibres provide clues to their function. For example, in The main projection of DG the molecular layer of the DG and the stratum lacuno- granular cells to CA3; sum-moleculare of CA3, projections from the EC con- characterized by high verge on to the apical dendrites of dentate principal cells II III V/VI II III V/VI concentrations of zinc. and interneurons. Specifically, the lEA projects to the LEA MEA PrS EC Schaffer collaterals outer third of the molecular layer of the DG, and the mEA PHR The axon collaterals of the CA3 projects to the middle third of this layer. A similar pat- PaS pyramidal cells that project to tern of convergence53 is observed in CA3, where the lEA CA1. projection terminates in the superficial part of the stratum lacunosum-moleculare and the mEA projection PER POR terminates in the deep part of this layer. In addition to convergence, divergence53 of the EC projections to the DG and CA3 also occurs, as individual layer II cells Neocortex project to both the DG and CA3 (REFS 48,54). Figure 3 | The standard view of parahippocampal– The organization of the EC projection to CA1 and hippocampal circuitry. The standardNature Re viewviews that | Neur is oscience the subiculum is markedly different from that of the presented here is based on various circuitry models from EC-to-DG or EC-to-CA3 projection. The origin of recent articles8–15. According to this standard view, the main projection from the EC to the stratum lacuno- neocortical projections are aimed at the parahippocampal sum-moleculare of CA1 and the molecular layer of the region (PHR), which in turn provides the main source of subiculum lies in layer III although, again, other layers input to the hippocampal formation (HF). In the PHR, two (II, V and VI) contribute to a lesser extent to this projec- parallel projection streams are discerned: the perirhinal tion20,22,25,26,29,38,41–43,46,49,51,52,55–57. Another striking feature of cortex (PER) projects to the lateral entorhinal cortex (LEA), this pathway is the difference between the lEA and mEA and the postrhinal cortex (POR) projects to the medial projections along the transverse axis. The lEA projects entorhinal cortex (MEA). The entorhinal cortex (EC) reciprocates the connections from the PER and the POR. to the distal part of CA1 and the proximal subiculum, Additionally, the EC receives input from the presubiculum whereas the mEA projects to the proximal part of CA1 (PrS). The EC is the source of the perforant pathway, which 38,49,52 and the distal subiculum . This segregation suggests projects to all subregions of the hippocampal formation. that the input from the lEA and the mEA is processed Entorhinal layer II projects to the dentate gyrus (DG) and in different parts of CA1 and the subiculum. This idea CA3, whereas layer III projects to CA1 and the subiculum is supported by the observation that the segregation of (Sub). The polysynaptic pathway, an extended version of the the EC input to CA1 and the subiculum is maintained traditional trisynaptic pathway, describes a unidirectional in the intra-HF projection from CA1 to the subiculum route that connects all subregions of the HF sequentially. In (see next subsection). short, the DG granule cells give rise to the mossy fibre In addition to this topology along the transverse axis pathway, which targets CA3. The CA3 Schaffer collaterals project to CA1 and, lastly, CA1 projects to the Sub. Output of the HF, there is a topological organization of connec- from the HF arises in CA1 and the Sub and is directed to the tions between the dorsolateral–ventromedial axis of the PHR, in particular to the deep layers of the EC. The Roman EC and the longitudinal axis of the HF: the dorsolateral numerals indicate cortical layers. parts of the lEA and the mEA project to the septal HF, the intermediate part of the EC projects to intermediate septotemporal levels, and the ventromedial EC projects in Supplementary information S4 (figure)). Another to the temporal HF40,58,59. According to some reports, the example of underexposed circuitry is the direct projec- actual organization of the perforant pathway is more tion from the PER and the POR to the HF. Both A35 widespread (see figure 6 in Supplementary information and A36 have been reported to project to CA1 and the S4 (figure)), such that this topography relates to the dens- subiculum23,62. The POR has been suggested to project est projections, whereas weaker components show a more to all sub-areas of the HF23, but another report indicates divergent distribution along the septotemporal axis46,50. only direct projections to CA1 and the subiculum57. Such a broader projection pattern along the septotemporal axis of the HF may affect information processing. Connectivity within the HF. In the standard model the The EC-to-HF projection forms the main PHR con- first step of the polysynaptic HF pathway (FIG. 3; see also nection to the HF. Other PHR subregions have also been figure 8a in Supplementary information S4 (figure)) is observed to project to the HF directly, although less formed by a unidirectional projection from the DG to strongly than the EC and most of them are not included CA3: the mossy fibres. The Schaffer collaterals, which origin the standard view. Neurons in all layers of the pre- inate in CA3 and project to CA1, are the next step in the subiculum and the parasubiculum project to the stratum polysynaptic loop. A detailed look at these connections moleculare of the DG32,44,60 and the subiculum24,32,60,61 shows an interesting topology along the transverse axis. and to the stratum lacunosum-moleculare of CA3 The distal part of CA3 projects to proximal CA1 and, (REFS 32,44) and CA1 (REFS 32,44,60) (see figures 7a and 7b conversely, the proximal part of CA3 projects to distal

CA1 (REFS 63–65). The topography of the projections 10a in Supplementary information S4 (figure)) and the that arise from mid-proximodistal portions of CA3 lies distal part of CA1 projects to the lEA49,52 (see figure between that of these two projection patterns. The last 10b in Supplementary information S4 (figure)). The step in the polysynaptic pathway is the projection from subiculum-to-EC projections have a similar topography CA1 to the subiculum. The proximal part of the CA1 along the long89,91 and transverse axes49,88,89,91,92, although pyramidal cell layer projects to the distal subiculum, they seem to be less sharply defined. moreover, along whereas the distal CA1 projects to the proximal part of the transverse axis the organization is opposite to that the subiculum52,66–69. of the CA1-to-EC connections: the proximal subiculum In contrast to what is depicted in the standard model, sends a stronger projection to the lEA and the distal there are several backprojections in the HF. Pyramidal subiculum sends a stronger projection to the mEA, cells in CA3 project back to the hilus and the inner again in line with the overall organization of the EC molecular layer of the DG64,70–74, and all septotem- projections to the subiculum. poral levels have this backprojection (see figure 8b in Although the CA1/subiculum-to-EC projections Supplementary information S4 (figure)). The strongest form the main part of the HF output to the PHR, other backprojection originates in the temporal levels of CA3 connections to the PHR also exist. For example, CA3 and projects to the temporal part of the DG71. Again con- (REFS 24,44,72,78), CA1 (REFS 24,31,44,69,75) and the trasting the standard idea of unidirectionality, a back- subiculum24,31,32,88,89,91,92 all project to the presubiculum projection from CA1 to CA3 has also been reported; this and the parasubiculum (see figure 11 in Supplementary backprojection most likely arises from inhibitory neu- information S4 (figure)). The projection from the subic- rons in the stratum radiatum and stratum oriens of CA1 ulum to the presubiculum is the best described of these. and projects to the same layers in CA3 (REFS 64,66,67,75) It follows a septotemporal gradient, such that the septal (see figure 8b in Supplementary information S4 (figure)). part of the subiculum projects to the septal presubicu- A backprojection from the subiculum to CA1 has also lum31,88,89,91 and the temporal part of the subiculum been reported (see figure 8b in Supplementary infor- projects to the temporal presubiculum24,91. A projection mation S4 (figure)). This backprojection arises from from the subiculum to the parasubiculum exists, but no neurons in the stratum pyramidale of the subiculum detailed information about it is known24,32,89. Finally, CA1 and projects to all layers of CA1 (REFS 32,76). Currently, and the subiculum project to both the PER and the POR, it is not known whether this backprojection is of an although no detailed information about the organization excitatory or an inhibitory nature. of this projection is currently available21,23. Recurrent collaterals of the CA3 region63,64,70–73,77–81 are well acknowledged in the literature (FIG. 3), and they have Functional implications been described in the other HF subregions as well (see In the preceding section we compared the details of the figure 9 in Supplementary information S4 (figure); these PHR–HF circuitry to the standard view, highlighting intrinsic recurrent networks are less extensive and are several underexposed connections. To provide a func- also less investigated in terms of their anatomy and functional perspective on some of these connections, we tion (see the ‘Functional implications’ section). In the now discuss them in the context of three topics that have polymorphic layer of the DG, each granule cell estab- long been associated with the HF: memory formation, lishes contact with the proximal dendrites of several navigation and temporal dynamics. mossy cells, which return excitatory synapses to granule cell dendrites in the molecular layer47,64,65,70,80,82–85. CA1 Memory formation. The first example of how increased has recurrent loops that are restricted to one septotem- knowledge of connections in the PHR and the HF might poral level66,69,75,76,79,86. In the subiculum, principal cells change our views on the memory function of the HF con- extend axon collaterals to a substantial part of the subicu- cerns the idea that the HF is the region in which different lum that lies ventral to the site of origin; these collaterals types of information are associated in memory. By con- terminate on pyramidal cells and interneurons32,76,87,88. trast, the EC is generally defined as a simple input–output structure that keeps the incoming information flows HF projections to the PHR. The HF output to the PHR separate by way of two parallel pathways (FIG. 3): the PER- arises from CA1 and the subiculum and, according to to-lEA-to-HF pathway conveys non-spatial information the standard view, terminates primarily in the deep lay- about external stimuli, whereas the POR-to-mEA-to-HF ers of the EC. In contrast to this view, several authors pathway conveys spatial information18. have reported direct projections from CA1 (REFS 72,75) However, there are four arguments that support the and the subiculum32,89,90 to the superficial layers of both notion that, rather than being a simple input–output the lEA and the mEA. structure, the EC has a role in more complex associa- There are reciprocal connections between the EC tions. First, anatomical evidence shows that PER and and CA1/the subiculum. The CA1–to–EC projection is POR projections to the EC overlap (see the ‘Circuitry’ organized such that the septotemporal axis of the HF section). Second, there is an extensive network in the EC is mapped topologically on to the dorsolateral–ventro- that reciprocally connects the lEA and the mEA17,20,29,37. medial axis of the EC, comparable to the organization of These first two anatomical characteristics suggest that the strongest EC-to-HF projection52,72,75. The transverse non-spatial information in the lEA and spatial informa- output organization also mimics the input — that is, the tion in the mEA can become associated at the level of the proximal part of CA1 projects to the mEA (see figure EC, which is supported by the observation that the lEA

is involved in odour–place associations93. Third, deep and the transverse axis, as the mEA and the lEA project pref- superficial layers of the EC are also anatomically intercon- erentially to different proximodistal regions of CA1 (see nected20,26,29,37,94, and this connection is likely to explain the the ‘PHR projections to the HF’ subsection). Preliminary observation that the firing characteristics of cells in all lay- data from recordings in the septal CA1 are in line with this ers of the mEA have a clear correlation across layers dur- idea and show that cells at different transverse positions ing the performance of spatial tasks95. Fourth, according have different firing characteristics101 that are related to the to the classical view (FIG. 3), the superficial EC layers are type of information provided by the mEA or lEA inputs the input layers to the HF, whereas the deep layers receive (spatial and non-spatial, respectively). We propose that hippocampally processed information that they convey CA1 is divided into subdomains along the combination of back to the cortex. However, the anatomical data summa- the septotemporal and proximodistal axes, and that each rized in this Review show projections from the deep layers subdomain independently processes different, specific of the EC to the HF, consistent with the finding that acti- combinations of information originating from different vation of the deep layers of the EC is sufficient to activate input areas. In theory, each of these subdomains would the DG96. Additionally, the HF projects to both deep and thus be able to encode and store unique input patterns, superficial EC layers (see the ‘Circuitry’ section). which may be instrumental in discriminating subtle dif- We therefore propose that the notion that the EC is ferences in input cues and may aid pattern separation and a simple, laminated input–output structure needs revi- completion. This prediction awaits further experimental sion: information becomes integrated before it enters the data, such as detailed recordings along both axes in freely HF. This suggests that both the HF and the EC associ- behaving animals. ate information that is relevant to memory. As the same types of information are processed by the two structures, Navigation. Different types of spatial information, dis- the question remains how their functions compare. One cussed below, are represented in the PHR–HF circuitry, way to view the distinctive roles of the regions is that the and the circuitry may facilitate the exchange of these dif- EC holds a more universal memory representation of ferent types of information in order to make navigation the associated information, whereas the HF is involved through an environment possible. The same circuitry in processing details of this information through may mediate the formation of memories for the spatial processes such as pattern separation and pattern com- position of behaviourally relevant cues. Place cells, which pletion. The observation that activity in the HF increases encode place fields, provide essential information for when a person is recalling details from memory supports navigation. They are found in CA1 (REF. 102) and CA3 the proposal that the HF has a role in processing detailed (REF. 103), but cells with similar functional properties information97. The idea that the EC processes informa- have been found in the subiculum104–106, the septal pre- tion at an earlier and more generic level than the HF (in subiculum107 and the parasubiculum108,109. In the HF, the which detailed information is processed) corresponds size of a place field is related to the septotemporal posi- to the idea that the EC holds a universal map that is tion of the place cells: place cells in the septal HF have the important to the HF in navigation, as discussed below. smallest place fields, at intermediate septotemporal lev- Associative networks and, in particular, the auto- els place fields are twice as big110 and in the temporal HF associative network of CA3 have been proposed to be they become even larger103,110,111. Place field size can be essential for encoding and storing episodic memories98,99. interpreted as a measure of spatial scale, indicating that The recurrent connections in this area can be theoretically environments might be represented at different spatial arranged into a number of discrete patterns of activation, resolutions along the septotemporal axis of the HF. called stable states or attractors, and the synaptic strengths A large number of non-overlapping, unique spatial of the recurrent connections determine the stable states of representations of the environment are stored in the rather this network98. Incoming information presumably directs limited network of the HF, which creates a storage prob- the network into one of its stable states, thus enabling lem. It has been argued that in order to solve this problem pattern completion100. Although the CA3 recurrent net- the HF might make use of a universal map, presumably work is currently thought to be the most elaborate in the located outside the HF102,112,113, that can be applied across HF, CA1, the DG hilus region and the subiculum also environments. Based on the strong reciprocal connectiv- contain recurrent collateral networks (see figure 9 in ity between the EC and the HF, the EC (in particular the Supplementary information S4 (figure)) and are likely to mEA) was considered a likely candidate for the location exhibit computational characteristics comparable to those of this map, as this area was shown to receive predomi- 16 Auto-associative network of the CA3 recurrent network. One striking feature of the nantly visuospatial information from the POR . Indeed, A network of neurons with CA1 recurrent network that emerges from the diagram a disruption of the monosynaptic information flow from axon collaterals that terminate is that the recurrent loops are restricted to one septotem- mEA layer III to CA1 affected long-term spatial-memory on dendrites of the parent cell. poral level (see figure 9 in Supplementary information performance114 and impaired place cell firing in CA1 Place cells S4 (figure)). For example, the input to the septal CA1 (REF. 115). However, initial recordings in the EC did not Principal neurons in the from CA3 arises from both septal and intermediate levels reveal cells with a striking spatially modulated firing pat- hippocampus and of CA3, whereas the input to the temporal CA1 arises tern116,117, probably because these recordings did not cover parahippocampus that fire from the temporal and intermediate CA3. This input is the most dorsolateral portion of the mEA. The dorso- whenever an animal is in a specific location in an then processed independently in both the septal and the lateral mEA was predicted to contain such cells because environment (corresponding to temporal CA1. It would be interesting to know whether it is reciprocally connected both to the septal hippocam- the cell’s ‘place field’). there is also regional specificity of CA1 recurrents along pus (see the ‘Circuitry’ section), in which place cells are

most conspicuous, and to visuospatial cortical domains by spatial information than the firing of cells in the — for example, the POR17,18. Subsequent recordings in the distal CA1 (REF. 101). A similar type of prediction can dorso lateral part of the mEA indeed revealed grid cells118. be made for the subiculum, as the lEA projects to the like place cells, grid cells show a gradual increase in grid proximal part of the subiculum and the mEA projects field size from the dorsolateral mEA towards the ventro- to its distal part. On the basis of this topology, the most medial mEA119 and, because of the predominant topology prominent place cells are expected to be found in the of the perforant path, the grid cells with the smallest grid distal subiculum. One study found subtle differences in field scale in the dorsolateral mEA connect to the place the spatial properties of cells in the proximal versus the cells in the septal HF with the smallest place field scale. distal subiculum104. There are several explanations for Similarly, the grid cells with the largest grid field scale in why the difference was only small, but the best expla- the ventromedial mEA connect to the place cells in the nation is probably the extensive but underexposed and temporal HF with the largest place field scale. not very well studied intrinsic recurrent network in the Head-direction cells are a third class of cells involved subiculum130. in navigation. Head-direction cells were first discovered in the septal presubiculum109,120, but directionally tuned Temporal dynamics. Some of the underexposed cells have also been observed in the EC95, the anterior PHR–HF connections are likely to be involved in the and lateral dorsal thalamic nuclei121–123, the lateral mam- temporal synchronization of neuronal firing between millary nucleus124, the retrosplenial cortex125 and the brain areas. Synchronized firing is essential for the striatum126. This indicates that the directional signal is coordination of spatially distributed networks and is probably computed in brain networks outside the HF. generally achieved through neuronal oscillations. By The head-direction information from the mammillary synchronizing excitatory periods across regions, oscil- bodies is crucial for place and grid cell functioning124, lations may facilitate the transfer of information in the and head-direction information from the presubiculum PHR–HF network131. Furthermore, oscillations pro- is important, although not indispensable, for the func- mote coincident firing among cells, which is likely to tional characteristics of place fields in CA1 (REF. 127). As be important for inducing synaptic plasticity (for exam- Grid cells the septal presubiculum also projects to other HF sub- ple, see REF. 132) and memory consolidation133. One Neurons in the entorhinal cortex that fire strongly when regions, we propose that the firing properties of neu- of the prerequisites for the occurrence of oscillations an animal is at one of several rons in the DG, CA3 and the subiculum might also be is the interaction between excitatory glutamatergic neu- specific locations in an affected by presubiculum lesions. rons and inhibitory GABA (γ-aminobutyric acid)-ergic environment and that are What more can the details of the circuitry tell us interneurons134. Different classes of GABAergic neu- organized in a grid-like fashion. about the space-related functional properties of the net- rons can be characterized in the hippocampus accord- Head-direction cells work? A first hypothesis is that information from the ing to their distinct firing patterns during behaviourally Neurons that fire only when head-direction system may enter the HF through at least relevant oscillations such as theta oscillations, gamma the animal’s head points in a two different routes. One route projects from the pre- oscillations and sharp wave ripples135–138; projections from specific direction in an subiculum directly to the HF and a second route runs these interneurons to different targets synchronize the environment. indirectly to the HF through the projections from layers firing of large numbers of pyramidal cells135. Although Mammillary bodies II and III of the EC. In order to decide which of these most research on GABAergic cells is carried out on A pair of nuclei in the routes provides the predominant directional input to the interneurons that project locally in one sub-area, recent hypothalamus, strongly HF, the reported effects of presubiculum lesions on CA1 evidence showed the existence of long-range GABAergic connected to the HF and the place cell firing107 should be compared with the effect of projection neurons that cross the sub-area border and anterior complex of the thalamus, that are involved in presubiculum lesions on the spatial-firing properties are involved in the coordination of spike timing across 139 recognition memory. of mEA neurons. If mEA neuron firing is not affected by sub-areas . such lesions, the direct route from the presubiculum to Although most tract-tracing studies do not reveal Theta oscillations CA1 is more likely to be the predominant input pathway whether a projection is excitatory or inhibitory, an indica- Rhythmical changes at 5–12 Hz in network activity, as for directional information to the HF. However, if the tion of the excitatory or inhibitory nature of a connection observed in the electro- firing properties of mEA neurons do change as a result can be derived from the layers of origin and termination. encephalogram, characteristic of presubiculum lesions, the CA1 firing properties after For example, the CA1-to-CA3 backprojections discussed of the hippocampal network a presubiculum lesion should be compared with the CA1 in the ‘Circuitry’ section arise not from the (excitatory) communicating with various firing properties after a selective mEA lesion115 and after glutamatergic principal cell layer, but mainly from neurons cortical and subcortical (REFS 64,66,67,75) networks in the brain. a combined presubiculum and mEA lesion. located in the stratum oriens of CA1 , Another prediction based on the PHR–HF network and project to the stratum radiatum and stratum oriens Gamma oscillations characteristics is that the place-specific firing of CA1 of CA3. An in vivo labelling study showed the same locus of Rhythmical oscillations of should be stronger at its proximal end than at its distal origin and termination for CA1 GABAergic neurons 25–70 Hz observed in the (REFS 77,140–142) electroencephalogram. end, as the mEA preferentially projects to the proximal projecting to CA3 . Also, in the stratum portion of CA1 and place-specific firing in CA1 strongly radiatum of CA1, cells project to the DG and the subic- Ripple oscillations depends on the direct input from the mEA115,128,129. By ulum. GABAergic cells have been reported to reside in Short-lasting bursts of field contrast, the preferential lEA-to-CA1 projection pat- the CA1 stratum radiatum with axons that radiate to the oscillations (~140–200 Hz) in tern predicts that non-spatial information about exter- molecular layer of the DG and the subiculum143. the mammalian hippocampus and parahippocampus that nal stimuli is processed in the distal CA1. Preliminary Because the layer of origin of these CA1 neurons occur during rest or slow-wave data show that the firing of cells in the proximal (mEA- seems to be a reliable predictor of GABAergic connec- sleep. recipient) CA1 is indeed significantly more affected tions, it is likely that other projections that do not start

in the principal cell layer are also GABAergic. This can Although topological information is available in the be used to discover the existence of other inhibitory pro- interactive diagram for a large number of connections, jections. In the interactive diagram (see Supplementary increasing the knowledge base of PHR–HF connectiv- information S1 (figure)) one can observe that, in the hip- ity is an important requirement for future functional pocampus, cells in the hilus of the DG project to CA1. understanding of these regions. Although currently all There are also reports of projections from the HF to the connections in the interactive diagram are displayed as PHR that do not start in the principal cell layer of the HF: if they are of equal density, we aim for future versions cells in the molecular layer of the subiculum project of the diagram (which will be available on our website) to the parasubiculum, the presubiculum and the POR. to differentiate between strong and weak connec- moreover, cells in the stratum oriens and the stratum tions. unfortunately, connectional density is often not radiatum of CA1 and the molecular layer of the subicu- reported quantitatively in the anatomical literature, and lum project to the lEA and the mEA. We suggest that even when it is reported it is a subjective observation these connections indeed originate from long-range that is difficult to compare between studies. Second, we GABAergic neurons, and are capable of function- aim to incorporate in vivo and in vitro electrophysiologi- ally coupling the PHR–HF subregions and coordinate cal data into future versions of this knowledge base so oscillations over the entire PHR–HF network. that it will contain information about the excitatory or inhibitory role of connections. Third, the current ver- Conclusions and future directions sion of the diagram displays only the layers of origin and Comprehensive knowledge of the organization of the termination, but each region and layer consists of several PHR–HF connectivity is of pivotal importance for elu- cell types. We aim for future versions of the diagram to cidating PHR–HF function. Such detailed knowledge of contain a description of pre- and postsynaptic cell types. PHR–HF circuits will help us to understand how these Implementing these improvements requires extensive circuits are engaged in spatial processing and temporal fundamental research into the cytoarchitectonic and dynamics, as well as in other functions that have been connectional properties of the region, but this invest- associated with the region, such as episodic memory144, ment will have a tremendous impact on advancing our crossmodal memory145, recollection and recognition146, functional understanding. memory for the temporal order of events147–150 and visual An ever-increasing amount of anatomical knowledge perception of conjunctions151. moreover, the PHR and HF brings with it several difficulties. One consequence of are implicated in various disorders, such as Alzheimer’s the overwhelming number of reported connections is disease152, epilepsy153, schizophrenia154 and depression155. that attention focuses on a selection of the connections Knowing the changes in connection patterns within and whereas others fall into disuse, especially those for which between these regions may help us to understand the the functional relevance is not entirely clear, such as underlying mechanisms of these PHR–HF-related disor- some of the recurrent collaterals in the HF subregions. A ders and consequently enhance the possibilities for treat- knowledge base such as the one presented in this Review ing them. This Review and the complementary knowledge can help to prevent the loss of valuable knowledge and base may facilitate the study of altered connectivity in inspire creative minds to come up with new solutions for animal models for diseases that involve the PHR and HF. outstanding problems in the field.

1. Ramón y Cajal, S. Estructura del asta de Ammon y 9. Brown, M. W. & Aggleton, J. P. Recognition memory: 16. Burwell, R. D. & Amaral, D. G. Cortical afferents of the fascia dentata. Anales de la Sociedad Española de what are the roles of the perirhinal cortex and perirhinal, postrhinal, and entorhinal cortices of the Historia Natural 22, 53–126 (1893). hippocampus? Nature Rev. Neurosci. 2, 51–61 rat. J. Comp. Neurol. 398, 179–205 (1998). 2. Scoville, W. B. & Milner, B. Loss of recent memory (2001). 17. Burwell, R. D. & Amaral, D. G. Perirhinal and after bilateral hippocampal lesions. J. Neurol. 10. Bussey, T. J. & Saksida, L. M. Memory, perception, postrhinal cortices of the rat: interconnectivity and Neurosurg. Psychiatry 20, 11–21 (1957). and the ventral visual-perirhinal-hippocampal stream: connections with the entorhinal cortex. J. Comp. 3. Zola-Morgan, S., Squire, L. R., Amaral, D. G. & Suzuki, thinking outside of the boxes. Hippocampus 17, Neurol. 391, 293–321 (1998). W. A. Lesions of perirhinal and parahippocampal 898–908 (2007). 18. Naber, P. A., Caballero-Bleda, M., Jorritsma-Byham, B. cortex that spare the amygdala and hippocampal This paper provides a critical but stimulating & Witter, M. P. Parallel input to the hippocampal formation produce severe memory impairment. perspective on the idea that the medial temporal memory system through peri- and postrhinal cortices. J. Neurosci. 9, 4355–4370 (1989). lobe system is involved only in memory, with special Neuroreport 8, 2617–2621 (1997). 4. Crick, F. & Koch, C. A framework for consciousness. emphasis on the functions of the perirhinal cortex. 19. Swanson, L. W. & Kohler, C. Anatomical evidence for Nature Neurosci. 6, 119–126 (2003). 11. Eichenbaum, H. A cortical-hippocampal system for direct projections from the entorhinal area to the 5. Amaral, D. G. & Lavenex, P. in The Hippocampus Book declarative memory. Nature Rev. Neurosci. 1, 41–50 entire cortical mantle in the rat. J. Neurosci. 6, (eds Andersen, P., Morris, R., Amaral, D. G., Bliss, T. & (2000). 3010–3023 (1986). O’Keefe, J.) 37–114 (Oxford Univ. Press, New York, 12. Eichenbaum, H., Yonelinas, A. P. & Ranganath, C. The 20. Kohler, C. Intrinsic connections of the retrohippocampal 2007). medial temporal lobe and recognition memory. Annu. region in the rat brain. II. The medial entorhinal area. 6. Witter, M. P. & Amaral, D. G. in The Rat Nervous Rev. Neurosci. 30, 123–152 (2007). J. Comp. Neurol. 246, 149–169 (1986). System 3rd edn (ed. Paxinos, G.) 637–703 (Elsevier This paper reviews current concepts of how 21. Deacon, T. W., Eichenbaum, H., Rosenberg, P. & Academic Press, San Diego, 2004). familiarity and recollection memory are supported Eckmann, K. W. Afferent connections of the perirhinal This book chapter gives an extensive description of by regions in the medial temporal lobe and cortex in the rat. J. Comp. Neurol. 220, 168–190 cells and projections in the HF and the PHR. proposes that there is a distinction between (1983). 7. Burwell, R. D. & Witter, M. P. in The Parahippocampal recollection and familiarity in specific regions of the 22. Kerr, K. M., Agster, K. L., Furtak, S. C. & Burwell, R. D. Region: Organization and Role in Cognitive function medial temporal lobe. Functional neuroanatomy of the parahippocampal (eds Witter, M. P. & Wouterlood, F. G.) 35–60 (Oxford 13. Hasselmo, M. E., Fransen, E., Dickson, C. & Alonso, region: the lateral and medial entorhinal areas. Univ. Press, New York, 2002). A. A. Computational modeling of entorhinal cortex. Hippocampus 17, 697–708 (2007). 8. Bird, C. M. & Burgess, N. The hippocampus and Ann. NY Acad. Sci. 911, 418–446 (2000). 23. Furtak, S. C., Wei, S. M., Agster, K. L. & Burwell, R. D. memory: insights from spatial processing. Nature Rev. 14. Martin, S. J. & Clark, R. E. The rodent hippocampus Functional neuroanatomy of the parahippocampal Neurosci. 9, 182–194 (2008). and spatial memory: from synapses to systems. Cell. region in the rat: the perirhinal and postrhinal This paper provides a comprehensive summary of Mol. Life Sci. 64, 401–431 (2007). cortices. Hippocampus 17, 709–722 (2007). arguments that support the thesis that studies on 15. Squire, L. R., Stark, C. E. & Clark, R. E. The medial 24. van Groen, T. & Wyss, J. M. The connections of the role of the rodent hippocampus in navigation temporal lobe. Annu. Rev. Neurosci. 27, 279–306 presubiculum and parasubiculum in the rat. Brain Res. bear on our understanding of memory. (2004). 518, 227–243 (1990).

25. Wyss, J. M. An autoradiographic study of the efferent 47. Segal, M. & Landis, S. Afferents to the hippocampus of subcortical, and bilateral hippocampal formation connections of the entorhinal cortex in the rat. the rat studied with the method of retrograde transport projections. J. Comp. Neurol. 302, 515–528 (1990). J. Comp. Neurol. 199, 495–512 (1981). of horseradish peroxidase. Brain Res. 78, 1–15 (1974). 70. Buckmaster, P. S., Strowbridge, B. W. & Schwartzkroin, 26. Lingenhohl, K. & Finch, D. M. Morphological 48. Steward, O. Topographic organization of the P. A. A comparison of rat hippocampal mossy cells and characterization of rat entorhinal neurons in vivo: projections from the entorhinal area to the CA3c pyramidal cells. J. Neurophysiol. 70, soma-dendritic structure and axonal domains. Exp. hippocampal formation of the rat. J. Comp. Neurol. 1281–1299 (1993). Brain Res. 84, 57–74 (1991). 167, 285–314 (1976). 71. Li, X. G., Somogyi, P., Ylinen, A. & Buzsaki, G. The 27. Honda, Y. & Ishizuka, N. Organization of connectivity 49. Tamamaki, N. & Nojyo, Y. Preservation of topography hippocampal CA3 network: an in vivo intracellular of the rat presubiculum: I. Efferent projections to the in the connections between the subiculum, field CA1, labeling study. J. Comp. Neurol. 339, 181–208 (1994). medial entorhinal cortex. J. Comp. Neurol. 473, and the entorhinal cortex in rats. J. Comp. Neurol. 72. Swanson, L. W., Wyss, J. M. & Cowan, W. M. An 463–484 (2004). 353, 379–390 (1995). autoradiographic study of the organization of 28. Van Haeften, T., Wouterlood, F. G., Jorritsma-Byham, B. 50. Tamamaki, N. Organization of the entorhinal projection intrahippocampal association pathways in the rat. & Witter, M. P. GABAergic presubicular projections to to the rat dentate gyrus revealed by Dil anterograde J. Comp. Neurol. 181, 681–715 (1978). the medial entorhinal cortex of the rat. J. Neurosci. labeling. Exp. Brain Res. 116, 250–258 (1997). 73. Wittner, L., Henze, D. A., Zaborszky, L. & Buzsaki, G. 17, 862–874 (1997). 51. Witter, M. P., Griffioen, A. W., Jorritsma-Byham, B. & Hippocampal CA3 pyramidal cells selectively innervate 29. Kohler, C. Intrinsic connections of the retrohippocampal Krijnen, J. L. Entorhinal projections to the aspiny interneurons. Eur. J. Neurosci. 24, 1286–1298 region in the rat brain: III. The lateral entorhinal area. hippocampal CA1 region in the rat: an underestimated (2006). J. Comp. Neurol. 271, 208–228 (1988). pathway. Neurosci. Lett. 85, 193–198 (1988). 74. Wittner, L., Henze, D. A., Zaborszky, L. & Buzsaki, G. 30. Segal, M. Afferents to the entorhinal cortex of the rat 52. Naber, P. A., Lopes da Silva, F. H. & Witter, M. P. Three-dimensional reconstruction of the axon arbor of studied by the method of retrograde transport of Reciprocal connections between the entorhinal cortex a CA3 pyramidal cell recorded and filled in vivo. Brain horseradish peroxidase. Exp. Neurol. 57, 750–765 and hippocampal fields CA1 and the subiculum are in Struct. funct. 212, 75–83 (2007). (1977). register with the projections from CA1 to the 75. Cenquizca, L. A. & Swanson, L. W. Spatial organization 31. van Groen, T. & Wyss, J. M. The postsubicular cortex subiculum. Hippocampus 11, 99–104 (2001). of direct hippocampal field CA1 axonal projections to in the rat: characterization of the fourth region of the 53. Sporns, O. & Tononi, G. in Handbook of Brain the rest of the cerebral cortex. Brain Res. Rev. 56, subicular cortex and its connections. Brain Res. 529, Connectivity (eds Jirsa, V. K. & McIntosh, A. R.) 1–26 (2007). 165–177 (1990). 117–147 (Springer, Berlin, 2007). 76. Finch, D. M., Nowlin, N. L. & Babb, T. L. 32. Kohler, C. Intrinsic projections of the retrohippocampal In this book chapter the authors clearly explain Demonstration of axonal projections of neurons in the region in the rat brain. I. The subicular complex. segregation and integration and describe a method rat hippocampus and subiculum by intracellular J. Comp. Neurol. 236, 504–522 (1985). for quantifying structural connection patterns as injection of HRP. Brain Res. 271, 201–216 (1983). 33. Caballero-Bleda, M. & Witter, M. P. Regional and global measures of brain dynamics. 77. Sik, A., Tamamaki, N. & Freund, T. F. Complete axon laminar organization of projections from the 54. Tamamaki, N. & Nojyo, Y. Projection of the entorhinal arborization of a single CA3 pyramidal cell in the rat presubiculum and parasubiculum to the entorhinal layer II neurons in the rat as revealed by intracellular hippocampus, and its relationship with postsynaptic cortex: an anterograde tracing study in the rat. pressure-injection of neurobiotin. Hippocampus 3, parvalbumin-containing interneurons. Eur. J. Neurosci. J. Comp. Neurol. 328, 115–129 (1993). 471–480 (1993). 5, 1719–1728 (1993). 34. Honda, Y., Umitsu, Y. & Ishizuka, N. Organization of 55. Braak, H. On the structure of the human archicortex. 78. Siddiqui, A. H. & Joseph, S. A. CA3 axonal sprouting connectivity of the rat presubiculum: II. Associational I. The cornu ammonis. A Golgi and in kainate-induced chronic epilepsy. Brain Res. 1066, and commissural connections. J. Comp. Neurol. 506, pigmentarchitectonic study. Cell Tissue Res. 152, 129–146 (2005). 640–658 (2008). 349–383 (1974). 79. Hjorth-Simonsen, A. Some intrinsic connections of the 35. Van Haeften, T., Wouterlood, F. G. & Witter, M. P. 56. Honda, Y., Umitsu, Y. & Ishizuka, N. Topographic hippocampus in the rat: an experimental analysis. Presubicular input to the dendrites of layer-V projections of perforant path from entorhinal area to J. Comp. Neurol. 147, 145–161 (1973). entorhinal neurons in the rat. Ann. NY Acad. Sci. 911, CA1 and subiculum in the rat. Neurosci. Res. 24 80. Van, G. T. & Wyss, J. M. Species differences in 471–473 (2000). (Suppl.), S101 (2000). hippocampal commissural connections: studies in rat, 36. Kohler, C., Shipley, M. T., Srebro, B. & Harkmark, W. 57. Naber, P. A., Witter, M. P. & Lopes da Silva, F. H. guinea pig, rabbit, and cat. J. Comp. Neurol. 267, Some retrohippocampal afferents to the entorhinal Evidence for a direct projection from the postrhinal 322–334 (1988). cortex. Cells of origin as studied by the HRP method in cortex to the subiculum in the rat. Hippocampus 11, 81. Lorente de Nó, R. Studies on the structure of the the rat and mouse. Neurosci. Lett. 10, 115–120 (1978). 105–117 (2001). cerebral cortex. II. Continuation of the study of the 37. Dolorfo, C. L. & Amaral, D. G. Entorhinal cortex of the 58. Witter, M. P. & Groenewegen, H. J. Laminar origin and ammonic system. J. Psychol. Neurol. 46, 113–177 rat: organization of intrinsic connections. J. Comp. septotemporal distribution of entorhinal and (1934). Neurol. 398, 49–82 (1998). perirhinal projections to the hippocampus in the cat. 82. Claiborne, B. J., Amaral, D. G. & Cowan, W. M. A light 38. Baks-Te-Bulte, L., Wouterlood, F. G., Vinkenoog, M. & J. Comp. Neurol. 224, 371–385 (1984). and electron microscopic analysis of the mossy fibers Witter, M. P. Entorhinal projections terminate onto 59. Witter, M. P. A survey of the anatomy of the of the rat dentate gyrus. J. Comp. Neurol. 246, principal neurons and interneurons in the subiculum: a hippocampal formation, with emphasis on the 435–458 (1986). quantitative electron microscopical analysis in the rat. septotemporal organization of its intrinsic and extrinsic 83. Frotscher, M., Seress, L., Schwerdtfeger, W. K. & Buhl, E. Neuroscience 136, 729–739 (2005). connections. Adv. Exp. Med. Biol. 203, 67–82 (1986). The mossy cells of the fascia dentata: a comparative 39. Deller, T., Martinez, A., Nitsch, R. & Frotscher, M. 60. Witter, M. P., Holtrop, R. & van de Loosdrecht, A. A. study of their fine structure and synaptic connections A novel entorhinal projection to the rat dentate gyrus: Direct projections from the periallocortical subicular in rodents and primates. J. Comp. Neurol. 312, direct innervation of proximal dendrites and cell complex to the fascia dentata in the rat. Neurosci. 145–163 (1991). bodies of granule cells and GABAergic neurons. Res. Commun. 2, 61–68 (1988). 84. Deller, T., Nitsch, R. & Frotscher, M. Heterogeneity of J. Neurosci. 16, 3322–3333 (1996). 61. Beckstead, R. M. Afferent connections of the the commissural projection to the rat dentate gyrus: a This paper described an underexposed projection entorhinal area in the rat as demonstrated by Phaseolus vulgaris leucoagglutinin tracing study. from the deep layers of the EC to the DG, which retrograde cell-labeling with horseradish peroxidase. Neuroscience 75, 111–121 (1996). should fundamentally alter our concept of the EC Brain Res. 152, 249–264 (1978). 85. Swanson, L. W. & Cowan, W. M. An autoradiographic (see also reference 96). 62. Kosel, K. C., Van Hoesen, G. W. & Rosene, D. L. study of the organization of the efferent connections of 40. Dolorfo, C. L. & Amaral, D. G. Entorhinal cortex of the A direct projection from the perirhinal cortex (area 35) the hippocampal formation in the rat. J. Comp. rat: topographic organization of the cells of origin of to the subiculum in the rat. Brain Res. 269, 347–351 Neurol. 172, 49–84 (1977). the perforant path projection to the dentate gyrus. (1983). 86. Finch, D. M. & Babb, T. L. Demonstration of caudally J. Comp. Neurol. 398, 25–48 (1998). 63. Ishizuka, N., Weber, J. & Amaral, D. G. Organization of directed hippocampal efferents in the rat by 41. Hjorth-Simonsen, A. Projection of the lateral part of intrahippocampal projections originating from CA3 intracellular injection of horseradish peroxidase. Brain the entorhinal area to the hippocampus and fascia pyramidal cells in the rat. J. Comp. Neurol. 295, Res. 214, 405–410 (1981). dentata. J. Comp. Neurol. 146, 219–232 (1972). 580–623 (1990). 87. Harris, E., Witter, M. P., Weinstein, G. & Stewart, M. 42. Kajiwara, R. et al. Convergence of entorhinal and CA3 64. Laurberg, S. Commissural and intrinsic connections of Intrinsic connectivity of the rat subiculum: I. Dendritic inputs onto pyramidal neurons and interneurons in the rat hippocampus. J. Comp. Neurol. 184, morphology and patterns of axonal arborization by hippocampal area CA1 - An anatomical study in the 685–708 (1979). pyramidal neurons. J. Comp. Neurol. 435, 490–505 rat. Hippocampus 18, 266–280 (2007). 65. Laurberg, S. & Sorensen, K. E. Associational and (2001). 43. Kohler, C. A projection from the deep layers of the commissural collaterals of neurons in the hippocampal 88. Witter, M. P., Ostendorf, R. H. & Groenewegen, H. J. entorhinal area to the hippocampal formation in the formation (hilus fasciae dentatae and subfield CA3). Heterogeneity in the dorsal subiculum of the rat. rat brain. Neurosci. Lett. 56, 13–19 (1985). Brain Res. 212, 287–300 (1981). Distinct neuronal zones project to different cortical 44. Nafstad, P. H. An electron microscope study on the 66. Amaral, D. G., Dolorfo, C. & varez-Royo, P. Organization and subcortical targets. Eur. J. Neurosci. 2, 718–725 termination of the perforant path fibres in the of CA1 projections to the subiculum: a PHA-L analysis (1990). hippocampus and the fascia dentata. Z. Zellforsch. in the rat. Hippocampus 1, 415–435 (1991). 89. Kloosterman, F., Witter, M. P. & Van Haeften, T. Mikrosk. Anat. 76, 532–542 (1967). 67. Swanson, L. W., Sawchenko, P. E. & Cowan, W. M. Topographical and laminar organization of subicular 45. Ruth, R. E., Collier, T. J. & Routtenberg, A. Topography Evidence for collateral projections by neurons in projections to the parahippocampal region of the rat. between the entorhinal cortex and the dentate Ammon’s horn, the dentate gyrus, and the subiculum: J. Comp. Neurol. 455, 156–171 (2003). septotemporal axis in rats: I. Medial and intermediate a multiple retrograde labeling study in the rat. 90. Van Haeften, T., Jorritsma-Byham, B. & Witter, M. P. entorhinal projecting cells. J. Comp. Neurol. 209, J. Neurosci. 1, 548–559 (1981). Quantitative morphological analysis of subicular 69–78 (1982). 68. Tamamaki, N. & Nojyo, Y. Disposition of the slab-like terminals in the rat entorhinal cortex. Hippocampus 46. Ruth, R. E., Collier, T. J. & Routtenberg, A. modules formed by axon branches originating from 5, 452–459 (1995). Topographical relationship between the entorhinal single CA1 pyramidal neurons in the rat hippocampus. 91. Naber, P. A. & Witter, M. P. Subicular efferents are cortex and the septotemporal axis of the dentate J. Comp. Neurol. 291, 509–519 (1990). organized mostly as parallel projections: a double- gyrus in rats: II. Cells projecting from lateral entorhinal 69. van Groen, T. & Wyss, J. M. Extrinsic projections from labeling, retrograde-tracing study in the rat. J. Comp. subdivisions. J. Comp. Neurol. 270, 506–516 (1988). area CA1 of the rat hippocampus: olfactory, cortical, Neurol. 393, 284–297 (1998).

92. Honda, Y., Umitsu, Y. & Ishizuka, N. Efferent 116. Quirk, G. J., Muller, R. U., Kubie, J. L. & Ranck, J. B. Jr. This paper provides evidence that particular projections of the subiculum to the retrohippocampal The positional firing properties of medial entorhinal long-range GABAergic projection neurons are and retrosplenial cortices of the rat. Neurosci. Res. 22 neurons: description and comparison with hippocampal important for coordinating oscillation across brain (Suppl.), S265 (1999). place cells. J. Neurosci. 12, 1945–1963 (1992). areas. 93. Mayeaux, D. J. & Johnston, R. E. Discrimination of 117. Frank, L. M., Brown, E. N. & Wilson, M. Trajectory 140. Gulyas, A. I., Hajos, N., Katona, I. & Freund, T. F. social odors and their locations: role of lateral encoding in the hippocampus and entorhinal cortex. Interneurons are the local targets of hippocampal entorhinal area. Physiol. Behav. 82, 653–662 (2004). Neuron 27, 169–178 (2000). inhibitory cells which project to the medial septum. 94. Van Haeften, T., Baks-Te-Bulte, L., Goede, P. H., 118. Fyhn, M., Molden, S., Witter, M. P., Moser, E. I. & Eur. J. Neurosci. 17, 1861–1872 (2003). Wouterlood, F. G. & Witter, M. P. Morphological and Moser, M. B. Spatial representation in the entorhinal 141. Losonczy, A., Zhang, L., Shigemoto, R., Somogyi, P. & numerical analysis of synaptic interactions between cortex. Science 305, 1258–1264 (2004). Nusser, Z. Cell type dependence and variability in the neurons in deep and superficial layers of the entorhinal 119. Brun, V. H. et al. Progressive increase in grid scale short-term plasticity of EPSCs in identified mouse cortex of the rat. Hippocampus 13, 943–952 (2003). from dorsal to ventral medial entorhinal cortex. hippocampal interneurones. J. Physiol. 542, 95. Sargolini, F. et al. Conjunctive representation of Hippocampus 18, 1200–1212 (2008). 193–210 (2002). position, direction, and velocity in entorhinal cortex. 120. Taube, J. S., Muller, R. U. & Ranck, J. B. Jr. Head- 142. Sik, A., Penttonen, M., Ylinen, A. & Buzsaki, G. Science 312, 758–762 (2006). direction cells recorded from the postsubiculum in Hippocampal CA1 interneurons: an in vivo intracellular This study provided evidence that information on freely moving rats. II. Effects of environmental labeling study. J. Neurosci. 15, 6651–6665 (1995). location, direction and distance is integrated and manipulations. J. Neurosci. 10, 436–447 (1990). 143. Vida, I., Halasy, K., Szinyei, C., Somogyi, P. & Buhl, updated in the grid cell network during navigation. 121. Mizumori, S. J. & Williams, J. D. Directionally selective E. H. Unitary IPSPs evoked by interneurons at the 96. Koganezawa, N. et al. Significance of the deep layers mnemonic properties of neurons in the lateral dorsal stratum radiatum-stratum lacunosum-moleculare of entorhinal cortex for transfer of both perirhinal and nucleus of the thalamus of rats. J. Neurosci. 13, border in the CA1 area of the rat hippocampus amygdala inputs to the hippocampus. Neurosci. Res. 4015–4028 (1993). in vitro. J. Physiol. 506, 755–773 (1998). 61, 172–181 (2008). 122. Taube, J. S. Head direction cells recorded in the 144. Tulving, E. in Organization of Memory (eds Tulving, E. 97. Viskontas, I. V., Carr, V. A., Engel, S. A. & Knowlton, anterior thalamic nuclei of freely moving rats. & Donaldson, W.) 381–403 (Academic, New York, 1972). B. J. The neural correlates of recollection: J. Neurosci. 15, 70–86 (1995). 145. Goulet, S. & Murray, E. A. Neural substrates of hippocampal activation declines as episodic memory 123. Stackman, R. W. & Taube, J. S. Firing properties of crossmodal association memory in monkeys: the fades. Hippocampus 19, 265–272 (2008). head direction cells in the rat anterior thalamic amygdala versus the anterior rhinal cortex. Behav. 98. Rolls, E. T. An attractor network in the hippocampus: nucleus: dependence on vestibular input. J. Neurosci. Neurosci. 115, 271–284 (2001). theory and neurophysiology. Learn. Mem. 14, 17, 4349–4358 (1997). 146. Mayes, A., Montaldi, D. & Migo, E. Associative 714–731 (2007). 124. Sharp, P. E. & Koester, K. Lesions of the mammillary memory and the medial temporal lobes. Trends Cogn. This paper provided a clear definition of the body region severely disrupt the cortical head Sci. 11, 126–135 (2007). characteristic features of attractor networks and direction, but not place cell signal. Hippocampus 18, 147. Manns, J. R., Howard, M. W. & Eichenbaum, H. Gradual summarized data in support of the view that the 766–784 (2008). changes in hippocampal activity support remembering hippocampus comprises an attractor network This study provides evidence that the mammillary the order of events. Neuron 56, 530–540 (2007). relevant for fast and efficient encoding and bodies provide important information for the 148. Fortin, N. J., Agster, K. L. & Eichenbaum, H. B. Critical retrieval of trial-unique memories. head-direction cells but not for the place cells in role of the hippocampus in memory for sequences of 99. Nakazawa, K. et al. Requirement for hippocampal CA3 the hippocampus. events. Nature Neurosci. 5, 458–462 (2002). NMDA receptors in associative memory recall. Science 125. Chen, L. L., Lin, L. H., Green, E. J., Barnes, C. A. & 149. Kesner, R. P., Gilbert, P. E. & Barua, L. A. The role of 297, 211–218 (2002). McNaughton, B. L. Head-direction cells in the rat the hippocampus in memory for the temporal order of 100. Gold, A. E. & Kesner, R. P. The role of the CA3 subregion posterior cortex. I. Anatomical distribution and a sequence of odors. Behav. Neurosci. 116, 286–290 of the dorsal hippocampus in spatial pattern completion behavioral modulation. Exp. Brain Res. 101, 8–23 (2002). in the rat. Hippocampus 15, 808–814 (2005). (1994). 150. Lisman, J. E. Relating hippocampal circuitry to function: 101. McNaughton, B. L. et al. Distinct characteristics of 126. Wiener, S. I. Spatial and behavioral correlates of recall of memory sequences by reciprocal dentate-CA3 CA1 place cells correlated with medial or lateral striatal neurons in rats performing a self-initiated interactions. Neuron 22, 233–242 (1999). entorhinal cortex layer III input. Society for navigation task. J. Neurosci. 13, 3802–3817 (1993). 151. Bussey, T. J., Saksida, L. M. & Murray, E. A. The Neuroscience Abstract 391.3 (2008). 127. Calton, J. L. et al. Hippocampal place cell instability perceptual-mnemonic/feature conjunction model of 102. O’Keefe, J. Place units in the hippocampus of the after lesions of the head direction cell network. perirhinal cortex function. Q. J. Exp. Psychol. B 58, freely moving rat. Exp. Neurol. 51, 78–109 (1976). J. Neurosci. 23, 9719–9731 (2003). 269–282 (2005). 103. Kjelstrup, K. B. et al. Finite scale of spatial 128. McNaughton, B. L., Barnes, C. A., Meltzer, J. & 152. Braak, H. & Braak, E. Neuropathological stageing of representation in the hippocampus. Science 321, Sutherland, R. J. Hippocampal granule cells are Alzheimer-related changes. Acta Neuropathol. 82, 140–143 (2008). necessary for normal spatial learning but not for 239–259 (1991). 104. Sharp, P. E. & Green, C. Spatial correlates of firing spatially-selective pyramidal cell discharge. Exp. Brain 153. McCormick, D. A. & Contreras, D. On the cellular and patterns of single cells in the subiculum of the freely Res. 76, 485–496 (1989). network bases of epileptic seizures. Annu. Rev. moving rat. J. Neurosci. 14, 2339–2356 (1994). 129. Brun, V. H. et al. Place cells and place recognition Physiol. 63, 815–846 (2001). 105. Sharp, P. E. Subicular cells generate similar spatial maintained by direct entorhinal-hippocampal circuitry. 154. Honea, R., Crow, T. J., Passingham, D. & Mackay, C. E. firing patterns in two geometrically and visually Science 296, 2243–2246 (2002). Regional deficits in brain volume in schizophrenia: a distinctive environments: comparison with hippocampal 130. Witter, M. P. Connections of the subiculum of the rat: meta-analysis of voxel-based morphometry studies. place cells. Behav. Brain Res. 85, 71–92 (1997). topography in relation to columnar and laminar Am. J. Psychiatry 162, 2233–2245 (2005). 106. Sharp, P. E. Subicular place cells expand or contract organization. Behav. Brain Res. 174, 251–264 155. Campbell, S. & MacQueen, G. The role of the their spatial firing pattern to fit the size of the (2006). hippocampus in the pathophysiology of major depression. environment in an open field but not in the presence 131. Buzsaki, G. & Draguhn, A. Neuronal oscillations in J. Psychiatry Neurosci. 29, 417–426 (2004). of barriers: comparison with hippocampal place cells. cortical networks. Science 304, 1926–1929 (2004). 156. Zaborszky, L., Wouterlood, F. G. & Lanciego, J. L. Behav. Neurosci. 113, 643–662 (1999). 132. Levy, W. B. & Steward, O. Synapses as associative Neuroanatomical Tract-Tracing: Molecules, Neurons, 107. Sharp, P. E. Multiple spatial/behavioral correlates for memory elements in the hippocampal formation. and Systems (Springer, 2006). cells in the rat postsubiculum: multiple regression Brain Res. 175, 233–245 (1979). This book provides excellent guidance on analysis and comparison to other hippocampal areas. 133. Wilson, M. A. & McNaughton, B. L. Reactivation of tract-tracing methods. Cereb. Cortex 6, 238–259 (1996). hippocampal ensemble memories during sleep. 157. Wickersham, I. R., Finke, S., Conzelmann, K. K. & 108. O’Keefe, J. in The Hippocampus Book (eds Andersen, P., Science 265, 676–679 (1994). Callaway, E. M. Retrograde neuronal tracing with a Morris, R., Amaral, D. G., Bliss, T. & O’Keefe, J.) 475– 134. Mann, E. O. & Paulsen, O. Role of GABAergic deletion-mutant rabies virus. Nature Methods 4, 579 (Oxford Univ. Press, New York, 2007). inhibition in hippocampal network oscillations. Trends 47–49 (2007). 109. Taube, J. S. Place cells recorded in the parasubiculum Neurosci. 30, 343–349 (2007). 158. Luo, L., Callaway, E. M. & Svoboda, K. Genetic dissection of freely moving rats. Hippocampus 5, 569–583 (1995). This paper reviews the mechanism by which of neural circuits. Neuron 57, 634–660 (2008). 110. Jung, M. W., Wiener, S. I. & McNaughton, B. L. GABAergic interneurons control the spike timing of This paper showed that a genetic approach can be Comparison of spatial firing characteristics of units in other neurons and synchronize network activity. an additional tool, next to anatomical and dorsal and ventral hippocampus of the rat. 135. Cobb, S. R., Buhl, E. H., Halasy, K., Paulsen, O. & physiological methods, for constructing circuit J. Neurosci. 14, 7347–7356 (1994). Somogyi, P. Synchronization of neuronal activity in diagrams. 111. Maurer, A. P., Vanrhoads, S. R., Sutherland, G. R., Lipa, P. hippocampus by individual GABAergic interneurons. & McNaughton, B. L. Self-motion and the origin of Nature 378, 75–78 (1995). Acknowledgements differential spatial scaling along the septo-temporal axis 136. Klausberger, T. et al. Brain-state- and cell-type-specific The authors would like to thank E. Moser and L. Colgin for of the hippocampus. Hippocampus 15, 841–852 (2005). firing of hippocampal interneurons in vivo. Nature providing helpful comments on an earlier version of this arti- 112. Touretzky, D. S. & Redish, A. D. Theory of rodent 421, 844–848 (2003). cle. N.L.M.C. acknowledges a personal grant from the navigation based on interacting representations of 137. Klausberger, T. et al. Spike timing of dendrite-targeting Netherlands Organization for scientific research (NWO) — space. Hippocampus 6, 247–270 (1996). bistratified cells during hippocampal network grant number 903-47-074. 113. Sharp, P. E. Complimentary roles for hippocampal oscillations in vivo. Nature Neurosci. 7, 41–47 versus subicular/entorhinal place cells in coding place, (2004). context, and events. Hippocampus 9, 432–443 (1999). 138. Tukker, J. J., Fuentealba, P., Hartwich, K., Somogyi, P. FUrtHer INFormAtIoN 114. Remondes, M. & Schuman, E. M. Role for a cortical & Klausberger, T. Cell type-specific tuning of hippocampal Authors’ homepage: www.temporal-lobe.com input to hippocampal area CA1 in the consolidation of interneuron firing during gamma oscillations in vivo. a long-term memory. Nature 431, 699–703 (2004). J. Neurosci. 27, 8184–8189 (2007). sUPPLemeNtAry INFormAtIoN See online article: S1 (figure) | S2 (box) | S3 (table) | S4 (box) 115. Brun, V. H. et al. Impaired spatial representation in 139. Jinno, S. et al. Neuronal diversity in GABAergic long- CA1 after lesion of direct input from entorhinal cortex. range projections from the hippocampus. J. Neurosci. all liNks are acTive iN The oNliNe pdf Neuron 57, 290–302 (2008). 27, 8790–8804 (2007).